Chatterbox
ElevenLabs-tier speech, MIT-licensed
What it is
A production-grade open TTS family: a 500M English model with unique emotion-exaggeration control, a 23-language multilingual variant with zero-shot voice cloning, and a 350M low-latency Turbo model with paralinguistic tags like [laugh]. Consistently benchmarks at or above closed competitors in blind listening tests.
Why it's interesting
A commercial voice-AI company giving away its crown jewels under MIT — including cloning — with built-in Perth audio watermarking as a responsible-AI hedge. The most-starred open TTS of the current cycle.
Use cases
- Voice agents and narration
- Dubbing across 23 languages
- Game dialogue and consented voice cloning
Who it's for
Developers shipping voice features, indie game devs, accessibility builders
Setup
Easy. pip install chatterbox-tts; a CUDA GPU is recommended but modest cards work, and Turbo is lighter
Limitations & cautions
Clone only consented voices. Reference-clip language should match the target to avoid accent bleed, and all output carries Resemble's watermark — a feature, but know it's there.
Editorial takeaway
The open-weights voice race ended faster than anyone expected. This is what winning it looks like.